A quick training done on still images (so don't expect crazy physics) to help with the appearance of erect penises.
It has its limits, it's a first jet. I'm still fiddling with training settings and will try and improve it over time. Consider it an alpha.
Best used without turbo lora as it might mess it up a bit but feel free to experiment.
Run at strength 1.0,
Trained using Ostris' AI Toolkit.
The audio might be a tiny bit fucked because Ai Toolkit loves to break the audio even when you don't train on audio. An advanced lora loader where the audio can be turned off might help.
Anyway, this is very early, the training adapter for Ai Toolkit is only in alpha, so let's be patient.
Description
V1.0
FAQ
Comments (21)
are your examples text to vid?
Yes, it's a T2V, as the title implies.
@DomDomTomTom ah lmao i knew i was misreading something. that's exciting! thanks for sharing
Best cover vid gen ever
Thanks. :3 you are great!
When I saw the thumbnail video, I instinctively flinched...
Then my job here is done.
me too, I love this lora.
Thanks a lot for this Lora.
Can the LoRA weight value be used to adjust penis length?
unlikely, that's the purpose of your prompt.
"mini" penis, "beef curtains", I'm beginning to think maybe most of these people using AI have never seen a naked person....
The model is called MINImax, Albert Einstein.
beef curtains :vomit:
Hahaha! They might use it even though they’ve never actually owned it.
how many images/steps in your training? i tried Ai Toolkit but i got only a potato penis
I've added a post to the gallery with a bracketed set of tests of the same prompt and seed with this LoRA at various strengths: 1.00, 0.75, 0.50, 0.30, 0.00 (control). Turbo "Larry" 600 step LoRA, 8 steps, full BF16 models.
i am getting some ugly penises with this lora, at 1.0 to .5 nothing works... i am using t2v
Yea.. I really wish the people that make the base models would stop being so selfish/ignorant and only inserting their own personal preferences (IE: female only, man don't exist) into models that they plan to distribute to thousands/millions of people.. Especially given that only literal millionaires can afford the hardware to train them..